In a workflow, each agent() call hands work to a separate agent, and Claude Code keeps a journal (a saved record) of each call's result so a stopped run can resume without redoing everything. Several things changed in that journal:
- Entries that were replaced or demoted are now tracked, and a resumed run rebuilds its chain from the journal when entries were replaced.
- Saved results are normalized into a consistent form before saving.
- If a result cannot be saved, a warning is logged and a resumed run will re-run from that call.
- When journaling is off, results that are not objects are refused.
- Agent state is now kept separately for each running context.
Resuming a workflow becomes more reliable, and when a result could not be saved you are warned that the run will repeat from that point instead of it failing silently.
tengu_workflow_schema_lint_enforce Not enough to sayNothing here resolved what this flag was doing on this version, so nothing here should be read as on or off.
This account: no value returned · anonymous baseline: no value returned · compiled default in v2.1.305: on
These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.
Read once, for one account on one subscription tier, against v2.1.305. It isn't a statement about your account. What a flag value here can and cannot tell you